Generative AI Utterance Resumption: A Unified Framework for Unidirectional and Bidirectional Expression Continuity Across Human–AI Communication

Generative artificial intelligence (AI) systems increasingly support extended interactions involving natural language generation, speech synthesis, real-time voice communication, and multimodal conversational activities. However, interruptions remain a fundamental challenge for maintaining continuous expression. When an AI system or human user pauses, suspends, or is interrupted during an unfinished utterance, the subsequent continuation may require repetition, reconstruction, contextual inference, or restarting the original expression. Existing research has investigated conversational interruptions, barge-in mechanisms, incremental speech generation, dialogue recovery, and speech playback control. Nevertheless, these research directions do not necessarily establish a unified conceptual framework in which both human users and generative AI systems are independently capable of preserving and resuming their unfinished expressions from their respective interruption positions. This paper introduces Generative AI Utterance Resumption (GAUR), a unified framework for preserving and restoring unfinished communicative expressions across human–AI interaction. The framework supports three principal communication configurations: AI-to-user unidirectional resumption, user-to-AI unidirectional resumption, and bidirectional human–AI resumption. It conceptualizes expression as a resumable communicative process rather than an indivisible conversational output. Its central mechanism preserves the expression state, interruption position, completed content, pending content, contextual dependencies, and continuation intent, enabling subsequent expression to resume from an appropriate preserved position. The framework further distinguishes playback-level, generation-level, and communicative-level resumption, establishing an architectural foundation for more comprehensive interruption-aware conversational systems. To the best of our knowledge, this paper presents the first unified conceptual framework specifically organizing human–AI utterance resumption across both unidirectional and bidirectional expression configurations. The contribution is theoretical and architectural; empirical validation and implementation are identified as future research directions.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-17
DOI
https://doi.org/10.5281/zenodo.24297655
Primary Topic
Speech and dialogue systems
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Generative AI Utterance Resumption: A Unified Framework for Unidirectional and Bidirectional Expression Continuity Across Human–AI Communication

Jincheng Zhang
15 links
Zenodo (CERN European Organization for Nuclear Research)
Speech and dialogue systems
preprint

Generative AI Utterance Resumption: A Unified Framework for Unidirectional and Bidirectional Expression Continuity Across Human–AI Communication

Jincheng Zhang
preprint en

Abstract

Generative artificial intelligence (AI) systems increasingly support extended interactions involving natural language generation, speech synthesis, real-time voice communication, and multimodal conversational activities. However, interruptions remain a fundamental challenge for maintaining continuous expression. When an AI system or human user pauses, suspends, or is interrupted during an unfinished utterance, the subsequent continuation may require repetition, reconstruction, contextual inference, or restarting the original expression. Existing research has investigated conversational interruptions, barge-in mechanisms, incremental speech generation, dialogue recovery, and speech playback control. Nevertheless, these research directions do not necessarily establish a unified conceptual framework in which both human users and generative AI systems are independently capable of preserving and resuming their unfinished expressions from their respective interruption positions. This paper introduces Generative AI Utterance Resumption (GAUR), a unified framework for preserving and restoring unfinished communicative expressions across human–AI interaction. The framework supports three principal communication configurations: AI-to-user unidirectional resumption, user-to-AI unidirectional resumption, and bidirectional human–AI resumption. It conceptualizes expression as a resumable communicative process rather than an indivisible conversational output. Its central mechanism preserves the expression state, interruption position, completed content, pending content, contextual dependencies, and continuation intent, enabling subsequent expression to resume from an appropriate preserved position. The framework further distinguishes playback-level, generation-level, and communicative-level resumption, establishing an architectural foundation for more comprehensive interruption-aware conversational systems. To the best of our knowledge, this paper presents the first unified conceptual framework specifically organizing human–AI utterance resumption across both unidirectional and bidirectional expression configurations. The contribution is theoretical and architectural; empirical validation and implementation are identified as future research directions.

Zenodo (CERN European Organization for Nuclear Research)
Speech and dialogue systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon